checkpoint-5/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 5,
  "global_step": 5,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 39.93763732910156,
      "learning_rate": 2e-05,
      "loss": 1.0503,
      "step": 1
    },
    {
      "epoch": 2.0,
      "grad_norm": 39.957481384277344,
      "learning_rate": 1.9749279121818235e-05,
      "loss": 1.0503,
      "step": 2
    },
    {
      "epoch": 3.0,
      "grad_norm": 9.691998481750488,
      "learning_rate": 1.900968867902419e-05,
      "loss": 0.901,
      "step": 3
    },
    {
      "epoch": 4.0,
      "grad_norm": 11.417628288269043,
      "learning_rate": 1.78183148246803e-05,
      "loss": 0.8168,
      "step": 4
    },
    {
      "epoch": 5.0,
      "grad_norm": 9.326613426208496,
      "learning_rate": 1.6234898018587336e-05,
      "loss": 0.7494,
      "step": 5
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.95,
      "eval_accuracy_<|content|>": 0.0,
      "eval_accuracy_<|from|>": -1,
      "eval_accuracy_<|recipient|>": -1,
      "eval_accuracy_<|stop|>": 0.0,
      "eval_accuracy_first_token": 1.0,
      "eval_accuracy_first_token_get": 1.0,
      "eval_accuracy_first_token_get_total": 2,
      "eval_accuracy_total_num_<|content|>": 2,
      "eval_accuracy_total_num_<|from|>": 0,
      "eval_accuracy_total_num_<|recipient|>": 0,
      "eval_accuracy_total_num_<|stop|>": 2,
      "eval_first_token_param_values": 0.8571428571428571,
      "eval_first_token_param_values_total": 14,
      "eval_loss": 0.6766262650489807,
      "eval_perplexity": 1.0099620212270384,
      "eval_runtime": 0.9535,
      "eval_samples_per_second": 1.049,
      "eval_steps_per_second": 1.049,
      "eval_total_number_first_token": 2,
      "step": 5
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 15,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 5,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1983832978882560.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}